Statistics 802 Quantitative Methods

advertisement
Final Thoughts
Goal (Syllabus)
To provide students
with a description of
the advanced
quantitative techniques
which are routinely
used for managerial
decision making
Goal (Syllabus)
 To provide students with examples of the application of
these models
 Interfaces
 Forecasting Project
 AHP Guest Lecture
Companies in Interfaces
presentations
Applying Quantitative Marketing Techniques to the Internet Structuring and sustaining
excellence in management science at Merrill Lynch People Skills: The need to change –
Problem or Opportunity? A new approach to performance management and goal setting
Against Your Better Judgement? How Organizations Can Improve Their Use of Management
Judgement in Forecasting Contract Optimization at the Texas Children's Hospital, Warner
Robins Air Logistics Center Streamlines. Aircraft Repair and Overhaul NBC-Universal
Uses a Novel Qualitative Forecasting Technique to Predict Advertising Demand. Integrating
Excel, Access and Visual Basic to Deploy Performance Measurement and Evaluation at the
American Red Cross. The “Killer Application” of Revenue Management: Harrah’s
Cherokee Casino & Hotel Optimally Stationing Army Forces Determinants of Success of
German Venture Capital Investments Investment Analysis and Budget Allocation at Catholic
Relief Services Spreadsheet Model Helps to Assign Medical Residents at the University of
Vermont's College of Medicine. Ranking US Army Generals of the 20th Century: A Group
Decision-Making Application of the Analytic Hierarchy ProcessDecision analysis; multiple
criteria; military; personnel. Designing the Response to an Anthrax Attack Managing Credit
Lines and Prices for Bank One Credit Cards Spreadsheet Models for Inventory Target Setting
at Procter & Gamble. Analyzing Supply Chains at HP Using Spreadsheet Models Supply
Chain Management: Technology, Globalization, and Policy at a Crossroads The Peoples Gas
Light and Coke Company Plans Gas Supply
Samples of Models
(From Lectures, Text, Homework, Greatest Hits and Exams)
 Market share
 Brand loyalty (Markov chain)
 Advertising (Game)
 Scheduling
 1 to 1 (Assignment)
 1 or many to many


Transportation
Integer Program (Set covering)
Samples of Models
 Advertising
 Media selection (linear programming)
 Competitive


Game/Market Share/$
Game/Price Guarantees – Guarantees guarantee
HIGH prices!
Samples of Models
 Inventory planning
 Newsboy problem (single period inventory model –
greeting cards example)


Decision table
Simulation
 Production planning - linear programming
 Bidding
 Simulation (in notes, we did not get to it except for one
team)
 Capital budgeting - integer program
Samples of Models
 Enrollment management/forecasting - Markov
chain
 Public services
 Mail delivery, street cleaning/plowing
 School bussing – transportation
 Finance/accounting
 Cost/volume - simulation
 Portfolio selection – linear/integer programming
Samples of Models
 Production
 Product mix/resource allocation - linear programming
 Blending - linear programming
 Employee scheduling- related problems
 Workforce scheduling
 Workforce training
 Assignment
 Health
 Diet problem
 Disease Progression
Samples of Models
 Location – game theory
 Agricultural planning
 Noncompetitive - linear programming
 Competitive - non zero sum game
Bonus Models - Sports
 Baseball
 Assignment of pitchers - linear programming
 States in a Markov chain
 Football
 Fourth and goal - decision tree
 Optimal sequential decisions and the content of the fourth-and goal
 Desperation - decision analysis - maximax
 Ice hockey
 Pull the goalie sooner
 Desperation - decision analysis - maximax
 Basketball
 Desperation - decision analysis - maximax
Models
In Some Cases There Is One Specific Goal
(maximize or minimize)
 Linear programming
 Transportation
 Assignment
 Integer programming
 Networks





Spanning tree
Shortest Path
Maximal Flow
Traveling Salesperson
Chinese Postman Problem
Models
In Other Cases There May Be More Than One
Specific Goal/Measurement
 Decision analysis
 Expected (monetary) value
 Maximin (conservative, pessimistic)
 Maximax (optimistic, desperate)
 Maximin regret (conservative, pessimistic)
 Forecasting
 Error measurement (technique evaluation)
 Mad
 Mean squared error (standard error)
 Mean absolute percent error (MAPE)
 Games
 Maximin/minimax
 Expected Value
Models
In Some Cases We are trying to rate or order
 Analytic Hierarchy Process (AHP)
 Data Envelopment Analysis
Prescriptive Vs.
Descriptive Models
 Some models PRESCRIBE what action to take
 Linear programming based
 Transportation, assignment, integer programming, goal
programming, game theory
 Network based
 Shortest path, maximal flow, minimum spanning tree, traveling
salesperson, Chinese postman
 Zero or constant sum games
 Flip a coin!!! –
Prescriptive Vs.
Descriptive Models
 Some models DESCRIBE the consequences of
actions taken
 Decision analysis
 Forecasting
 Markov chains
 Simulation
 Non zero sum games
 Matching lowest price leads to high prices !
 Competition leads to low prices
 Ranking


AHP
DEA
Probabilistic vs. Deterministic Models
 Some models include probabilities
 Markov Chains
 Decision Analysis


Decision tables
Decision trees
 Games
 Forecast Ranges
 Simulation
Probabilistic vs. Deterministic Models
 Other models are completely deterministic
 Linear programming



Transportation
Assignment
Data Envelopment Analysis
 Integer programming
 Networks
 AHP
Long Run
 Some models/measures require steady state (long run) in
order for the results to be useful
 Games
 Decision analysis


Expected value
Expected value of perfect information
Models
Tradeoffs
 Ease of use vs. flexibility/generality
 Transportation (easier) vs. LP (more flexible)
 Decision table (easier) vs. Decision tree (more
flexible)
 QM for windows (easier) vs. Excel (more
flexible)
 Model correctness vs. solvability
 Integer programming/linear programming
Models
Tradeoffs
 Model Exactness vs. Flexibility
 Analytical method vs. Simulation
 Development Cost/Time vs. Exactness
 Analytical method vs. Simulation
Model Sensitivity
 Forecasting & Simulation
 Standard error/standard deviation
 Linear Programming
 Dual values/ranging table
 Integer Programming
 Change values 1 unit at a time
 Decision Tables/Decision Trees
 Data table (letting probabilities vary)
Data Table With a Decision Tree
Solving Backwards
 Decision tree
 Game tree (sequential decisions)
 Let’s make a deal
Models –
Number of Decision Makers
 One
 Most models
 More than one
 Games

Let’s make a deal !!
 AHP – sort of
Excel Addins
 Solver
 Linear & integer programs
 Networks (shortest path & maximal flow)
 Zero sum games
 Crystal ball
 Simulation/risk analysis
 Will be used in your Fall Finance course
 Excel QM
 Decision trees
 Many other models
Excel Tools
 Data analysis
 Forecasting
 Simulation

Can be used for generating random numbers
 Scenarios
 Data tables
 Simulation
 Decision tables
 Decision trees
Computer Skills
 Microsoft office
 Word
 Excel
 Blackboard
 Discussion Board
 Listserv
 Software
 Download
 Installation
Less important computer skills (but skills
nonetheless)
 QM (POM-QM) for Windows
 Will be used in MSOM 5806 – Operations Mgt in Fall
 Change menu now
 Excel OM
 Available for use in MSOM 5806
 (requires new file rather than a menu change)
SURVEY/EVALUATION RESULTS
CLASS OF 2009
Survey Results – Forecasting
Class of 2009/2008/Class of 2007/Class of 2006
 Workload
 Too much time – 2/3/1/5
 Just right –
12/25/17/18
 Too little time – 2/1/0/0
 Value
 High –
13/22/18/17
 Medium – 2/6/1/6
 Low –
1/1/0/0
 Conclusion: Maintain project as is.
Interfaces presentations
 Workload
Conclusion:
 Too much time – 1/2/1/2
Continue,
but
 Just right –
15/26/18/20
consider students
 Too little time – 0/0/0/1
using ppt
 Value of reading; listening
 High –
5;5/12;10/10;6/7; 6
 Medium – 8/6;14;10/7;6/14; 11
 Low –
3;2/3;3/1;1/2; 1
 Interfaces options
 Discontinue –
2/3/2/17
 Continue as is–
6/10/10/1
 Continue w Power point – 5/12/10/na
LP interpretations
self
 Workload
 Too much time – 0/1/0/2
 Just right –
16/26/18/20
 Too little time – 0/2/0/0
 Value
 High –
11/10/13/14
 Medium – 5/10/6/8
 Low –
0/0/0/0
 Conclusion: Continue as is
LP interpretations
team
 Workload
 Too much time – 3/2/1/7
 Just right –
13/26/17/16
 Too little time – 0/1/0/0
 Value
 High –
10/10/11/12
 Medium – 5/17/5/8
 Low –
1/1/3/3
 Conclusion: Continue as is
Decision Tree - Team
 Workload
 Too much time – 1/3
 Just right –
14/23
 Too little time – 1/2
 Value
 High –
10/14
 Medium – 6/12
 Low –
0/3
 Conclusion: Continue as is
Decision Simulation
 Workload
 Too much time – 0
 Just right –
13
 Too little time – 3
 Value
 High –
6
 Medium – 8
 Low –
2
 Conclusion: Your evaluation results above say continue as
is but your performance indicates that I need to make it
more challenging
Group Take home exam
 Workload
 Too much time – 2/2/2/6
 Just right –
14/24/16/17
 Too little time – 0/3/0/0
 Value
 High – 13/22/16/21
 Medium – 3/7/3/2
 Low –
0/0/0/0
 Conclusion: Continue
Homework/Exam
 Workload
 Too much time – 6/5/2/14
 Just right –
9/18/12/8
 Too little time – 1/6/4/1
 Value
 High –
7/15/12/14
 Medium – 8/13/7/7
 Low –
1/1/0/2
 Conclusion: Continue as is
Guest Lecture
 Repeat next year – 10/18/13/13
 Do not repeat –
6/9/6/9
 Conclusion: Continue. Based on
comments I will ask Bob to dive
into the AHP program at 9:30.
Overall Course Workload
 Compared to Econ, Elective
 Above average – 10/13/7/15
 Average –
6/16/11/8
 Below average – 0/0/0/0
 Compared to Stat 5800
 Higher – 8/13/3/6
 Same – 7/14/14/16
 Lower – 1/2/1/1
 Conclusion: Workload may be slightly high
THE FINAL EXAM & GRADES
Final Exam
 Howard, now is the time to return the exams!
base = 108
Comparisons
Pct (Cl 08/07/06)
Mean
75
70% (76%, 75%, 71%)
Median
82
76% (78%, 79%, 74%)
Max
99
92% (100%, 95%)
 Exam this year had only 4 problems
 Note: Overall performance was best I have seen!
Student Grade Sheet
The End
Download